This post was featured on KevinMD.com; an excerpt follows.
"Full code" is the universal default status for patients who haven't chosen otherwise. Yet I suspect most physicians believe this policy is wrong. We feel in our hearts we're doing harm when we perform CPR on poor souls whose bodies are trying to naturally end their own suffering.
Appropriately timed end-of-life discussions are the supposed answer, but they're emotionally draining, interpersonally complex, and time-consuming. For a busy and stressed physician caring for an ill patient, it's easiest to either avoid the discussion entirely or stick to concrete verbal formulations, putting the patient or the family on the spot with ridiculous questions like "Do you want us to do everything?" ... "If your heart stops, should we restart it?" ... "If he stops breathing, should we put him on a breathing machine?"
One major source of the complexity is the irreducible uncertainty in the prognosticating of survival from in-hospital cardiac arrest (as an example, a majority of patients suffering ventricular tachycardia/fibrillation on the medical ward survive to discharge with good neurologic outcomes, while <2% of those with PEA arrests while on pressors in the ICU do). Hospitals' response has been to maintain the universal "full code" default, and the courts have avoided making any guiding rulings in the area. The result is an excessive burden on physicians by essentially requiring us to climb this difficult mountain with every patient, every family, every time.